AU #9 in ESPN's Football Power Index

[CleCoTiger 1,957]

I'm not going to pretend to understand the ESPN Football Power Index or the algorithms they use to derive it but after this weekend's games, AU is at #9 with UGA at #8 and bammer at #1.  

The win probabilities for AU at this time based on the FPI have our Tigers favored in three of the four remaining games with a 79.5% probability against aTm, 55.7% win probability (not much better than a coin toss) against UGA, 98.8% against La-Mo but only 28.2 percent against bammer.

(Jeff Sagarin's computer system also seems to still like our Tigers, ranking AU at #10.  Under Sagarin's current rankings, the UGA game would have the Dawgs a very slight favorite.)

None of this means anything of course. After all, the ESPN FPI had our Tigers with a very high probability of whipping LSWho (which I think they should have.)  Just tossing this into the discussion mix because I'd like to hear some input/opinions on the ESPN FPI.  I know in recent years it has had a decent track record at season's end. If it plays out as they have it now, our Tigers end the regular season a 9 win team with a chance to make it 10 in a bowl.

 

 

[AUld fAUx@ 2,584]

from a bit different fundamental perspective, a somewhat similar outlook:

“—  Predictably, Auburn cut through Arkansas like a hot knife through butter, winning 52-20. The Tigers have looked like a top-tier team this year. Their two losses (to Clemson and LSU) were by a combined 12 points. It’s not shaping up to be a championship-contending season for Gus Malzahn and Co., but this team will be dangerous come bowl season.”

[Jesse Reed at https://www.msn.com/en-us/sports/ncaafb/top-takeaways-from-college-football-week-8/ar-AAtQqdf?li=BBnb7Kz&ocid=mailsignout]

[djg0017 726]

I believe in FPI, but I think it tends to show a team’s true potential rather than measuring their greatness. I think playcalling brings our team way down from our potential. Aside from weird turnovers, injuries, bad plays, FPI seems to be a good projection model IMO. Unfortunately our team has lots of those things 

[weagl1 1,782]

The FPI must just somehow measure talent rather than coaching.  I agree we have top 10 talent but when you factor in coaching where we are in the polls now is about right.  We can pour in on against a some lesser talented teams but when the talent is more equal we come up short and when we are the lesser talented team we have no chance.  See past results against LSU, UGA, Clemson, Okie and Bammer.  

[TexasTiger 12,829]

1 hour ago, CleCoTiger said:

I'm not going to pretend to understand the ESPN Football Power Index or the algorithms they use to derive it but after this weekend's games, AU is at #9 with UGA at #8 and bammer at #1.  

The win probabilities for AU at this time based on the FPI have our Tigers favored in three of the four remaining games with a 79.5% probability against aTm, 55.7% win probability (not much better than a coin toss) against UGA, 98.8% against La-Mo but only 28.2 percent against bammer.

(Jeff Sagarin's computer system also seems to still like our Tigers, ranking AU at #10.  Under Sagarin's current rankings, the UGA game would have the Dawgs a very slight favorite.)

None of this means anything of course. After all, the ESPN FPI had our Tigers with a very high probability of whipping LSWho (which I think they should have.)  Just tossing this into the discussion mix because I'd like to hear some input/opinions on the ESPN FPI.  I know in recent years it has had a decent track record at season's end. If it plays out as they have it now, our Tigers end the regular season a 9 win team with a chance to make it 10 in a bowl.

 

 

It's hard for math to factor in gross coaching incompetence.

[AURealist 2,185]

I wonder if ESPNfpi gives us the edge over UGA because they've only played one hard game - ND and we play UGA @ home.

 

[CleCoTiger 1,957]

17 minutes ago, AURealist said:

I wonder if ESPNfpi gives us the edge over UGA because they've only played one hard game - ND and we play UGA @ home.

 

Dang if I know. The notes at the bottom of the page say that FPI "measures team's true strength on net points scale; expected point margin vs average opponent on neutral field."  Maybe there's more explanation of their methods/algorithms somewhere.  (I do know that in last weeks games with all the upsets in the top 10 playing against unranked opponents, the ESPN FPI was about as "accurate" with its predictions as I am week in and week out in the TD AU Bowl. LOL's!)

On edit...

ESPN had a blog post in 2015 explaining the FPI.  Reading it now.

[AURealist 2,185]

2 minutes ago, CleCoTiger said:

Dang if I know. The notes at the bottom of the page say that FPI " measures team's true strength on net points scale; expected point margin vs average opponent on neutral field."  Maybe there's more explanation of their methods/algorithms somewhere.  (I do know that in last weeks games with all the upsets in the top 10 playing against unranked opponents, the ESPN FPI was about as "accurate" with its predictions as I am week in and week out in the TD AU Bowl. LOL's!)

IIRC, they do multiple calculations to generate a team's expected-point-value per play and then add a custom QB rating into the mix and then adjust per game based on the likelihood of the 1st string QB playing, strength of schedule and home field advantage.  

Looking at things that way, if the basic formulas are good, the FPI should be increasingly accurate as the season progresses.  They do the same for the NFL.

[AURealist 2,185]

10 minutes ago, CleCoTiger said:

Spoiler

 

Dang if I know. The notes at the bottom of the page say that FPI "measures team's true strength on net points scale; expected point margin vs average opponent on neutral field."  Maybe there's more explanation of their methods/algorithms somewhere.  (I do know that in last weeks games with all the upsets in the top 10 playing against unranked opponents, the ESPN FPI was about as "accurate" with its predictions as I am week in and week out in the TD AU Bowl. LOL's!)

On edit...

 

ESPN had a blog post in 2015 explaining the FPI.  Reading it now.

Thanks for that.

[CleCoTiger 1,957]

5 minutes ago, AURealist said:

Thanks for that.

No problem. I'm following links from that post and going further down the rabbit hole.  This link to ThePredictionTracker.com comparing all the different systems and ratings is interesting as heck:  http://www.thepredictiontracker.com/ncaaresults.php?year=17

[AURealist 2,185]

3 minutes ago, CleCoTiger said:

No problem. I'm following links from that post and going further down the rabbit hole.  This link to ThePredictionTracker.com comparing all the different systems and ratings is interesting as heck:  http://www.thepredictiontracker.com/ncaaresults.php?year=17

Whoa!

I used to like TeamRankings before they changed format.

[CleCoTiger 1,957]

7 minutes ago, AURealist said:

Whoa!

I used to like TeamRankings before they changed format.

Other than ESPN's FPI, Sagarin, Dunkel and Massey, I have to say that I was pretty much unaware of most of the systems that are being tracked at ThePredictionTracker.com.  I had no idea.  (Heck, I still miss the old days of Leonard's Losers on the radio.  )

[OnthePlains 2,014]

College football is so unpredictable and so much is based on motivation, emotion, excitement etc.  If Auburn is motivated when the Dawgs come to JHS then I think its a fight to the finish.  If they lose to A&M and feel like the season is lost then UGA might come in and blast our doors off effectively ending the Gus Malzahn era at Auburn.  Either way, the FPI won't take those emotional factors into account. 

[wdefromtx 3,159]

I have a spreadsheet that compares how teams actually performed against the averages of their opponents and using that information creates a defensive and offensive multiplier to come up with a score for a match-up.  

 

I need to come up with a coefficient for the "Gus effect." LOL...........It is based on if the team is playing at it's statistical potential, meaning if we have a normal game for our team.  I exclude FCS games.

 

It has historically been pretty accurate since I started messing with it in 2013. It was predicting a 27-14 win against LSU and based on the stats we were well on our way to do better until Gus decided it was such a great idea to run the ball up the middle a million times. 

 

It is predicting we should win against TAMU about 30 to 20.

On the flip side we show to lose against the turds by a score of about 26 to 16.

 

I will update the TAMU game once they play MSST this weekend. 

[wdefromtx 3,159]

Also based on my methodology MSST should beat TAMU this weekend.

 

The average score I am getting is MSST 32 TAMU 25.

 

From a pure performance based standpoint against their competition MSST is a better team. But, TAMU seems to be trending up.